The background of the invention can be appreciated from the following literature:
[1] Betzig et al, Science 313, pp. 1642-1645 (2006);
[2] Hess et al., PNAS 104, pp. 17370-17375 (2007);
[3] Hess et al., Biophys J. 91, pp. 4258-427 (2006);
[4] Shroff et al., PNAS 104, pp. 20308-2031 (2007);
[5] Rust et al., Nature Methods 3, pp. 793-796 (2006);
[6] Egner et al., Biophys J. 93, pp. 3285-3290 (2007);
[7] Toprak et al., Nano Lett. 7, pp. 2043-2045 (2007);
[8] Juette et al., Nature Methods 5, p. 527 (2008); and
[9] Huang et al., Science 319, p. 810 (2008).
For some time now, various methods for overcoming the diffraction limit have been developed and applied in fluorescence microscopy (PALM, STED, Structured Illumination). A method for high-resolution fluorescence microscopy that is developing rapidly at present is based on highly precise localization of individual molecules. It is known that localization, that is, determination of the position of an individual fluorescent molecule, is not subject to diffraction limits (see references). This localization can be performed using wide-field high sensitivity cameras with a precision reaching into the nm range, if a sufficient number of photons of the molecule can be detected. In high-resolution microscopy based on localization, an image is composed from the molecule positions obtained in this manner. It is critical in this respect that only a subset of the molecules of the sample are in a fluorescent state, so that on average, the “nearest-neighbor” distance of the active molecules is always greater than the PSF of the microscope. This is achieved by using optically or chemically switchable fluorophores: in a densely marked region of a sample, stochastic subsets of fluorophores in the region of interest are switched by irradiation of a suitable conversion wavelength into the fluorescent state. The spot density is adjusted in such a way that the molecule positions can be localized continually. This optical switching method is used, for example, in Photo Activated Localization Microscopy (PALM). Variations of this fundamental method are described in detail in the literature [1-6].
Its variants (PALM, STORM, D-STORM, etc.) primarily differ in the selection of the fluorophores and the type of optical switching process.
However, all methods have in common that molecule localization is achieved by imaging on a highly sensitive camera (e.g. an EMCCD). The quasi-point light source (molecule) is represented by the point-spread function (PSF) of the microscope over several camera pixels. The precise position of the molecule on the x/y plane can now be determined either by fitting of the known PSF (Gaussian) or by determining the center of gravity, or by a combination of the two (Gaussian Mask Fit).
Typical localization precision ranges from 5 to 30 nm (depending on the experimental conditions); this then also represents approximately the lateral resolution of this method. The requirement that molecules should not be located too closely to one another on the one hand and that the examined structures should be represented as completely as possible, on the other, means in practice that many individual images (typically 20,000) must be taken of the sample. In each image, the positions of the molecules that are active at that point in time are determined and stored. Thus, considerably longer calculation or analysis times (depending on the algorithm and computer system used) are added to the already considerable image recording time for 20,000 images before the actual high-resolution image is available.
The method of high resolution based on localization described above is, however, limited to surfaces of two dimensions, since the localization of each dye molecule is incomparably more complex in the third spatial dimension (z-direction). Several approaches are known from the literature that will be briefly discussed below.
Astigmatism/Cylindrical Lens (See Reference [9]):
In this approach, a weak cylindrical lens is inserted into the detection-beam path, which leads to an astigmatic PSF. Accordingly, the image of the molecule is elliptically distorted when the molecule is located above or below the symmetry point of the PSF. Information can then be extracted about the z-position of the molecule from the orientation and magnitude of the distortion. A problem with this method is that the local environment and the orientation of the molecular dipole can also result in a distortion of the spot of the molecule (see above references). These molecules would then be assigned a false z-value depending on their orientation.
Detection on Two Planes:
In this technique, a 50/50 beam splitter is inserted into the detection beam path that splits (duplicates) the image into two partial images. These two images are either displayed on two identical cameras or side-by-side on a camera chip. An optical difference in path lengths is introduced into one of the two partial beam paths so that two object planes develop from the two partial beam paths, apart from each other in the z-direction by approximately half to one z-PSF (700 nm). The z-position for molecules located between these two planes can now be determined, e.g. by means of subtracting the two partial images of the molecule and/or by fitting a three-dimensional PSF.
For this method, either two highly sensitive cameras are required or both images must be arranged side-by-side on a camera chip. The latter will naturally result in a limitation of the image field. Furthermore, both variants require precise alignment of the beam paths or calibration measurements in order to ensure sub-pixel-precise overlapping of the two partial images.
The methods described above for localization in the x/y plane and/or in the z-direction can be used not only for high-resolution microscopes, but also for particle or molecule tracking, respectively. This applies accordingly to the potential solutions presented below.
Various Approaches are Known for Detection on Multiple Planes:
The following are examples:
Beam splitting and generation of two object planes through different image distances: See Bewersdorf et al. [8], Toprak et al. [7]
AT 402 863 B describes beam splitting in the detection using two cameras, at least one of which can be moved axially to change the object distance. The purpose is a comparative representation of images based at different object depths.
WO 95/00871 describes chromatic beam splitting on two detectors for 3D representation of objects.
U.S. Pat. No. 5,982,497 describes chromatic beam splitting and laterally offset imaging at the same object and image distance for representation of two or more color channels on one image sensor.
References [7] and [8] are most relevant for high-resolution depth localization of individual molecules in the z-direction.
However, the solutions described in the above references have the following disadvantages and limitations:                Oblique incidence on the image sensor causes distortions of the PSF depending on the z-position; the center of gravity is laterally dependent on the z-position;        The imaging scale changes as a function of the z-position;        Splitting cannot be adjusted, meaning that the optimal working point cannot be adjusted for different microscope objectives;        If an adjustment were (theoretically) made (such as by the mirror method described in [8]), it would be impossible to set the z-splitting on both planes to “zero,” due to a longer beam path in air;        An adjustment made in this way would shift the moved object plane into the cover glass at an increased z-splitting when measuring objects near the cover glass in high resolution, requiring refocusing of the entire system.        The problem to be overcome by the present invention therefore is splitting a microscope image into two partial images on two partial areas of an image sensor while avoiding the disadvantages and limitations mentioned above.        